Thesis

A comprehensive approach to ship system maintenance modelling and decision support using machine learning and reliability analysis

Downloadable Content

Download PDF
Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2024
Thesis identifier
  • T16907
Person Identifier (Local)
  • 201859833
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • Maintaining machinery health and repair data is essential for efficient maintenance planningand implementation. Identifying critical components and failure causes requires detailedsystem reliability and diagnostics analysis. Multi-equipment holdings and long voyages acrossmultiple climates make this challenge especially difficult for ship operators. For naval ships,the challenging environment and mission profile forced machinery to operate outside itsoperational envelope. Thus, this research seeks to develop a critical component analysismaintenance framework for system reliability and fault identification analysis to aidmaintenance decision-making. Using reliability analysis and machine learning, criticalcomponents and faults were identified. A unique contribution of this study is the integration offault detection analysis and reliability tools. DFT and FMECA are used to identify missioncritical components, while BBN is used for availability assessment and maintenance decisionsupport system. This includes classification and fault detection using ANN-based machinelearning models. An offshore patrol vessel power generation system with four marine dieselgenerators was studied. The reliability analysis shows system reliability below 70% in the first24 of 78 operational months. Over 40% of subsystem failure and related events were isolatedusing reliability importance measures and minimal cuts sets. Identifying mission-criticalcomponents using Risk Priority Number in FMECA analysis enabled robust reliability andcritical component analysis. Among the 4 MDGs, the lubricating system had the highestaverage availability of 67% and the cooling system the lowest at 38% using the DFTA minimalcut set. DSS-based 4 maintenance strategies used BBN availability and FMECA missioncritical components. Because some critical parts fail frequently, Corrective Action andConMon were recommended maintenance strategies. ANN found overheating when MDGoutput was above 180kva, linking component failure to generator performance. The findingsimprove ship system reliability and availability by reducing failures and improvingmaintenance strategies.
Advisor / supervisor
  • Lazakis, Iraklis
Resource Type
DOI
Funder

Relations

Items

Thumbnail Title Date Uploaded Visibility Actions
file details PDF of thesis T16907 2024-08-27 Public Download